Mapping Functionality to Data Domains

[Former Member]

Hi Experts

I hope you can guide me with regards to one of the Hybris Marketing concepts.

Different screens provide different perspectives on “customers” but it seems these “customers” are not all the same type of entity.

• Customers can be replicated from ECC using SLT. Out of the box Hybris Marketing provides a HANA attribute view of 17 tables all from schema (SAP_CUAN_ERP) that appears to be aliased to HMD_LTR. (I assume this schema name stands for Landscape Transformation from system HMD).
• Customers can be replicated from CRM also using SLT. Out of the box Hybris Marketing provides a HANA view joining 20 or so tables to a schema SAP_CUAN_CRM. My system lacks this schema but as yet we don't have CRM replication set up so that would be the reason.
• It is possible (I’m guessing here) that Hybris Marketing carries its own native set of tables to carry customer data derived from neither CRM nor ECC.We have a schema SAPHHA which might be the native Hybris Marketing one
• Segmentation can be performed on any custom tables providing HANA views are created and suitably aliased into Hybris Marketing.
• Sentiment  Engagement seems to be yet another set of tables. This can accept data from multiple sources and uses the email address as a key to unify these customers. It seems this is unrelated to the CRM and ECC customers.

As far as I can tell not all these “customers” can be accessed from the same screens. 

The table below shows the types of customer and the areas of functionality. At this point I am uncertain about many of these and would appreciate being corrected where I have the wrong understanding.

Area of Functionality	Audience Discovery	Customer Value Intelligence	Social Contact Interaction	Predictive Intelligence
Native Hybris Marketing	Yes	Yes	Yes	Yes
ERP replicated data	Yes	Yes		Yes
CRM Replicated Data	Yes	Yes
Custom Tables	Yes
Social Media Feeds			Yes

My next question is about the CRM and ECC data. If we have both systems which of the following scenarios are supported:

• Use SLT to bring in a feed from each of the two systems?
• Feed ECC data into CRM and then just bring a single SLT feed from CRM?
• Feed CRM data into ECC and then just bring a single SLT feed from ECC?

With regards to the sentiment engagement are the “customers” in this domain in any way linked to the ECC and CRM customers, or is this a totally independent piece of analysis?

The next concerns other new sources. I understand a client can bring in segmentation data and contact interaction data from almost any source. What about for customer value insight and some of the other functional areas. We have been told that Hybris Marketing provides a single view of the customer regardless of the source. If we wanted to bring in a third-party source could we then access it in the customer data management areas (e.g. customer engagement or margin analysis?) What are the steps required to do such an import?

There are out-of-the-box views to support predictive analytics for ECC and CRM. There is also a Banker Demo that uses runs predictive analytics over custom tables. It appears the process of building a predictive analytics over a new source is non-trivial. If we were to bring in new sources and load them to custom tables is it practical to build predictive analytics over them?

Normally basic questions of entity are addressed long before we get into detailed analysis of screen functionality. It seems I am missing a key area of knowledge. Is there a document that describes this?

Thanks

Cliff Stinson

[former_member183196 · Accepted solution]

Clifford,

First of all I want to appreciate your efforts into analyzing the hybris Marketing data model to full extent.   Now I can share somethings based on my experience. 

SAP hybris Marketing has 3 data models from which it powers all its application functionalities such as segmentation, campaign management..etc.  There are some good reasons why it is like that.  One of the key reason is to enable our own SAP customers to get going very easily and also support 3 party data sources.

     1. SAP ERP data model

     2. SAP CRM data model

     3. Interaction data model (meant for any 3rd party data but also subset of data from SAP ERP/CRM ends up here when certain reports are run) 

Segmentation:

By default we provide two segmentation profiles.  One based  ERP/CRM data model (aka customer segmentation which is good for B2B marketing) and the other based on interaction data model (aka contact segmentation which is good for B2C marketing).

If customers have all non-SAP data, then they load data interaction model using our rapid data load package or methods described in our inbound interface guide http://help.sap.com/mkt#section2

Also, they typically extend the interaction model by enhancing the "include" in the table CUAND_CE_IA_RT table.

If customers have only SAP data,  then they simply replicate using SLT and use  "customer segmentation".

If customers have both SAP and non-SAP data, then SLT the SAP data, then run a "CUAN_IC_MASTERDATA_EXTR_FULL/DELTA" report to extract customer data into interaction model as contacts.  Then they can use "contact segmentation".

Sentiment Engagement works on top interaction model (SOCIALDATA, and SOCIALUSERINFO tables play key role).  This way, even sentiment scoring is available in contact segmentation.

Customer Value Intelligence(now called "insight"):

Provides lot of KPI such as revenue margin, discount %...etc.  This is based on SAP ERP/CRM data model and more applicable for B2B scenarios.  If you are have customer that still want to use this but have non-SAP data, then they have to map their data into SAP ERP data model by leveraging one of the scope item from rapid data load package.  I can guess this large effort so I see this is less applicable scenario for non-SAP data.  This is a classic problem of data model rigidity vs flexibility with respect to out of box reporting. 

See below for your information

Data Model	Segmentation	Insight	Sentiment Engagement	Predictive Intelligence
Interaction data model	Yes	NA	NA	Yes
ERP replicated data	Yes	Yes	NA	Yes
CRM Replicated Data	Yes	Yes	NA	Yes
Custom Tables	Yes	probably using data mapping during load	NA	Yes but requires Dev
Social Media Feeds ( part of interaction model)	Yes	NA	Yes	NA

Question: If we were to bring in new sources and load them to custom tables is it practical to build predictive analytics over them?

Answer:  New source data need to be loaded as interactions.  Incase interactions can not accomdate all fields from sources, then extend it or join it to your customer tables via contact key.  Then, contact segmentation works.  At this point, you simply create your KPI using r-library or infinite insight then extend the contact segmentation.  This way, you can build your own predictive KPI on top of new sources. 

Hope I answered all your questions atleast at high level.  We are working on training course and will publish it in near future.

Thanks

-Bharath